Body mass index and mortality in a lean Chinese population

B91

Background:
 >High BMI is associated with increased risk of total mortality and with some specific causes of mortality. Standard cutoffs for overweight and obesity may not be suitable for all populations. Asian populations may demonstrate a different association between BMI and mortality. The population of Linxian China is lean and has very high rates of upper-gastrointestinal cancer, with stroke and heart disease being the other two main causes of death. We sought to examine the relationship between BMI and both total and cause-specific mortality in a long-term prospective cohort study conducted in Linxian.
 >Method:
 >The General Population Nutrition Intervention Trial cohort enrolled 29,584 adults aged 40-69 years from four Linxian communes in 1985, and height and weight were measured at that time. Subjects were actively followed, and causes of death recorded, from the beginning of the intervention trial in March 1986 through May 2001. We estimated the association between baseline BMI and subsequent mortality using age-adjusted Cox proportional hazards models.
 >Results:
 >Over 15 years of follow-up, 9686 people died, including 2974 deaths from stroke, 2705 from upper-gastrointestinal cancer, and 2005 from heart disease. BMI quintiles were defined as ≤20.0, 20.1-21.2, 21.3-22.3, 22.4-23.7, ≥23.8 and subjects in BMI quintile 4, which had the lowest mortality rate, were used as the reference group. Because few women had ever smoked cigarettes, we stratified the cohort for analysis into male smokers, male non-smokers, and females. The relative risks for all-cause mortality in male smokers for quintiles 1, 2, 3, and 5, compared to 4, were 1.27 (1.15-1.41), 1.01 (0.91-1.12), 0.94 (0.84-1.04), and 1.10 (0.97-1.24) respectively. In male non-smokers, relative risks were 1.11(0.95-1.31), 1.01 (0.86-1.19), 1.07 (0.92-1.25), and 1.22(1.04-1.42). In females they were 1.11 (1.01-1.21), 1.04 (0.94-1.15), 1.01 (0.91-1.11), and 1.12 (1.02-1.23). We found a direct association in all groups between BMI and stroke, such that higher BMI was associated with higher stroke risk throughout the distribution. For upper gastrointestinal cancer, male smokers and women were at lower risk with higher BMI, but in non-smoking men no significant association was found. Low BMI was associated with increased risk of heart disease death in each group. Further adjusting for alcohol intake, socioeconomic status, and some dietary factors did not materially alter these associations.
 >Conclusions:
 >We found the highest risks of total mortality occurred in persons in with the lowest and highest BMIs These associations differed by cause of death, with low BMI conferring increased risk for each of the outcomes except stroke. Although the association between BMI and mortality was generally similar to other populations, we did find some differences in this lean Asian population.